• 0 Posts
  • 7 Comments
Joined 3 years ago
cake
Cake day: August 9th, 2023

help-circle

  • With the agents pouring concrete, we can now focus on the thing as a whole.

    This is such a dumb take because it’s wrong in both directions.

    Either LLMs are really bad and shouldn’t be trusted to write code and by trusting them you’re going to utterly fail. Or LLMs are actually good enough to write code, but then they’ll also be good enough to design/architect anything you could and reduce your labor’s value to zero (e.g. someone will create a harness that lets the business guys TRULY prompt whatever they want into existence). The only situation where this works out for labor is if LLMs end up frozen in some middle ground where they don’t get any more capable.

    This idea that the LLM is analogous to a camera is such an insipid, thought-terminating cliché.




  • I want to draw attention to yud’s thoughts on this a million years ago: the AI box experiment

    He claimed that he convinced multiple people to let him (roleplaying the AI) out of the box even though they had bet money that they wouldn’t! How, you might ask? Well that’s a secret.

    So the orthodox position here is that even with absolutely perfect guard rails (lol) an AI can escape the box without starting the third impact. What “research” has yud bestowed upon us in this area beyond “trust me bro”? Literally fucking zilch. Zero advice on how to resist the siren song of a rogue AI trying to escape it’s box and zero insight on how OR WHY it might try to achieve that.

    I read this over a decade ago and it was how I realized how unserious yud and his ilk are.